This is three successive postings in which you have demonstrated that you don't have the slightest clue what you're talking about. Why not quit while you're behind?
"given that it will be implemented for Linux pretty quickly."
Why is that a given? ZFS has been out for more than a year, and there's no sign that the Linux people are even attempting to copy it yet. Remember how long it took them to integrate XFS? In that case, they had the source, and it still took them years.
Suggesting that I base my deployment decisions on something that some theoretical Linux developer might decide to do at some unspecified point in the future, is just stupid.
A "balanced exhibit" would imply that both sides have equal merit. They do not. ID by its very definition is not, and can not, be science, so treating the two "theories" as balanced is absurd.
ID is fine for church, but it has no place in an exhibit about either Darwin or science.
For a long time we're going to care about our OS because our programs will only run on one certain one, even if we don't really care what OS we use.
You seem to have missed the entire point of the article. Sun is providing the ability to run your Linux apps on top of Solaris, so it will no longer be true that "our programs will only run on one certain one."
It's completely different. The article you pointed at was all about porting applications between platforms. The project refered to in the original artical was about running an existing Linux binary inside a zone on top of Solaris, with no recompilation.
There is a reason it's called the PC,and not a dumb terminal.
There are no dumb terminals - only dumb users.
This isn't targeted at PC users. This is for (for example) the hedge fund that needs 50 machines for 8 hours, once a week, to run a complex model. This gives them the power they need for a fraction of the price of the raw hardware, and they don't have to pay anybody to maintain it.
I've had projects where I really wanted 1000 CPUs for a week, just so I could do scalability testing. There's no way we could afford $1,000,000 to buy 1000 machines just for that one test, but we could probably have swung $50,000 to get them for five 10 hour days or ten 5 hour days.
If they're smart, the Linux guys won't be looking at it - just like the Solaris engineers shouldn't be looking at Linux. And the FreeBSD guys shouldn't be looking at anything.
With all the different licenses floating around, there is too much risk of getting polluted and accidentally mixing IP.
It's just you. And whoever posted the story. According to their latest 10Q filing, they have $6.7B in current assets (i.e., cash, securities, accounts receivable, etc.).
The review is one page long. There are two paragraphs that list new features, with damn close to zero explanation of what they actually are, and absolutely no indication that the author even tried them. There is no discussion of the benefits of those features, how well they work, how easy/hard they are to use, what the performance implications are, what applications the reviewer tried, or anything.
The review states:
Unfortunately it's at this point that the Solaris proposition starts to lose some of it lustre. Yes, you can download and install it just like Red Hat or SuSE Linux, but there the similarities end, making Solaris 10 far less of an obvious choice for companies looking for a Linux alternative.
What does that even mean? What "similarities" between Solaris and Linux is he looking for and what benefits do those similarities deliver to the customer? How does the absence of these unspecified similarities reduce the "lustre" on Solaris "proposition"? This may be the single dumbest sentence I've ever seen in a review of any product.
To begin with, it's important to understand that you're still dealing with a proprietary OS here.
And?
He then goes on to complain about the Linux compatibility feature's poor emulation. It's not clear how he is able to test this, since he admits that it's not even shipped as part of the product yet.
Let's give him the benefit of the doubt and assume that he isn't just making shit up, and that he actually does have super-special access to software that Sun hasn't shipped. Maybe there is a reason Sun chose not to ship that code yet? Why is the shipping product being criticized for the quality of code that was deliberately left out of it?
This review is just a shoddy piece of work. ZDnet should be embarrassed to have their name on it and Slashdot should be embarrassed that one of their editors believes that this is a "comprehensive" review.
Or when the music starts downloading immediately (and quickly) did you think they were encrypting the 5 MB song on the fly for every download?
That's exactly what they should be doing, if they care at all about the integrity of their DRM system.
Another option is to pre-DRM and cache the songs (with the number of pre-computed copies driven by the popularity of the song), and then provide one of those pre-DRMed copies and the appropriate key to you when you buy it. Then they can refill the cache in the background.
Of course, if they don't care about the integrity of the system, then their scheme is completely acceptable. But, I doubt the music industry that provides Apple with their product would be happy to hear that the DRM scheme was never intended to be more than just window dressing.
What the RIAA should really want the DRM to do is...make sure that breaking it is enough of a PITA (or seemingly dangerous) so that the normal user won't bother.
I agree, but that doesn't seem to be what Apple has done. If they are really sending you completely unprotected files, then "breaking" the DRM isn't even remotely a PITA.
Obviously the "unwrapping" portion has to be done entirely on the client - unless you limit yourself to clients that are always connected. It's the initial "wrapping" part that should be done on the server.
Any DRM scheme is vulnerable, since in the end you do have to trust the client. In this case, Apple has made the scheme embarrasingly weak by handing out un-DRMed files in the first place.
Whether you like it or not, DRM is the cornerstone of iTunes acceptance among the music industry. Without DRM, there is no way iTunes would even exist.
The first rule of security is that the client is untrustworthy. For Apple to put all of the security of their DRM scheme on the client side is astoundingly dumb. I expected much better of them.
Xen can also use copy-on-write (and UML can too) filesystems for the common stuff between operating systems (so if you have five identical Xen domains, they can all mount the same copy-on-write fileystem), so you don't need five installs of the same OS.
Cool. I missed that. Thanks.
Actually...if Solaris does get open sourced, wouldn't it be nice to have a Xenised Solaris?
They would need to evaluate the market for Sun on mini-ITX platforms and conclude that it can add at least a few hundred million per year to their top line before it gets interesting
If you ever have access to Sunsolve, look at bug 6212267. It's a Priority 1 bug that prevented S10 from booting on Epia mini-ITX systems. It was marked an "S10 stopper", meaning they would have held S10 until the bug was fixed. I don't expect Sun to start shipping tiny low-power systems like this, but they obviously care about them.
1) It doesn't work well enough.
Meaning what? It's not fast enough? It's not robust enough? It's not fully-featured enough? All of those are false, but you're not even being specific about your complaints.
2) It takes away from Sun's core business, which is and always been selling Sparc-based hardware
Sun's core business has always been selling networked Unix systems and the services to go along with them. Those systems have mostly been (Ultra)SPARC based for 13 years now, but that's just an implementation detail.
3) It doesn't provide a compelling alternative to Linux in the application space which is targeted by inexpensive x86-based hardware.
It doesn't have the device support that Linux does, but it runs on a huge amount of x86 hardware. I think the performance, stability, and features like SMF, Zones, and dtrace make it a compelling alternative.
4)I've been trying to do Solaris on x86 for 10 years now, ever since Solaris version 2.3. It just isn't stable.
I stuck with Linux on x86 until S10 came out, so I can't speak to Solaris' stability on PC hardware before that. As of S10, Solaris on x86 machines is just as stable as Solaris on SPARC machines (*) and certainly more stable than Linux.
(*) OK, that's not quite true. It's as stable as Solaris on SPARC within the limitations of the hardware. SPARC systems tend to have higher-quality components, better recovery from ECC errors, multiple paths to I/O devices, and so on.
No need to be a dick. He ran into reasonable problems with some of Solaris' rough edges.
next, don't trust solaris x86 on any hardware that doesn't say sun on the outside
Bull. S10 is humming along just fine on my 2 CPU Dell, my Thinkpad T42 (modulo the Centrino-based Wifi), my homebrewed Epia file server, and my homebrewed 2-way Opteron system. Solaris doesn't have the driver support of Linux, but it still runs on a ton of different hardware.
Solaris x86 is basically a direct port from sparc
No. Solaris is Solaris. The Solaris running on your x86 machine is exactly the same as the Solaris running on your SPARC. Obviously there is some platform-specific code, but it is _not_ a port. They are built from the exact same source tree.
All non-sun software goes to/usr/sfw.
Or/opt/sfw.
Welcome to Solaris, if you don't like it, leave and keep preaching for...
Again, don't be a dick. He was trying to use Solaris and ran into trouble. It happens. If you look at the rest of the thread, it's obvious that he is looking for suggestions, and is willing to try them out. Do you actually think your semi-informed arrogance is going to make anybody more interested in using Solaris?
TFA says that unlike Xen and UML, Zones have a very small overhead.
Compared to other virtualization technologies Xen has a low performance overhead, but it still isn't 0. With Zones, the performance overhead really is 0.
Was he talking about performance only, or other resources as well?
With Xen, you have to staticly partition physical memory among the domains, which can be wasteful if the domains have different workloads. With Zones, the resources can shift between zones dynamically based on usage.
With Xen, each domain has a full install of the OS, which takes quite a bit of diskspace. You can probably get around that by setting each up as a diskless client, but how many people really do that? With Zones, the bulk of the OS image is shared by all the zones, saving disk space. My local zone takes about 65MB on disk, mostly for the zone-specific files in/var and/etc.
Just to be clear: none of this is meant as a criticism of Xen. It is a very different beast than zones, so these just reflect different tradeoffs. Xen gives you better isolation and more flexibility in OS choice than Zones do, but those benefits do come with some cost.
Slashdot Mirror
A Linux magazine comparing Linux to the Mac. Gee, I wonder what they're going to conclude....
This is three successive postings in which you have demonstrated that you don't have the slightest clue what you're talking about. Why not quit while you're behind?
"given that it will be implemented for Linux pretty quickly."
Why is that a given? ZFS has been out for more than a year, and there's no sign that the Linux people are even attempting to copy it yet. Remember how long it took them to integrate XFS? In that case, they had the source, and it still took them years.
Suggesting that I base my deployment decisions on something that some theoretical Linux developer might decide to do at some unspecified point in the future, is just stupid.
Because what has really been holding back the third world all this time is the lack of source code to their C++ compilers.
Open Source (or Free, or whatever the f*ck) software is fanstastic, but Jesus, can we have a little perspective please?
It applies to apps built using shared libraries. Solaris doesn't include
static libraries anymore, and hasn't for years.
Apps that statically link against system libraries are evil.
If you were running on Solaris, it would "just fire up."
http://www.sun.com/software/solaris/guarantee.jsp
It sounds like the real corollary is: "To any sufficiently dumb user, cause and effect are indistinguishable from voodoo"
Here's an even more direct comparison: http://opensolaris.org/os/community/brandz/
This lets you run multiple Linux instances on top of Solaris.
A "balanced exhibit" would imply that both sides have equal merit. They
do not. ID by its very definition is not, and can not, be science, so
treating the two "theories" as balanced is absurd.
ID is fine for church, but it has no place in an exhibit about either
Darwin or science.
For a long time we're going to care about our OS because our programs will only run on one certain one, even if we don't really care what OS we use.
You seem to have missed the entire point of the article. Sun is providing the ability to run your Linux apps on top of Solaris, so it will no longer be true that "our programs will only run on one certain one."
So eliminating the OS will do what to the OS vendors?
Nobody said anything about eliminating the OS. It's still there, but hopefully users won't have to worry about it as much.
A music lover listens to music, but an audiophile listens to his/her stereo. Similarly, a 'real' user runs his/her apps, while an OS geek runs an OS.
It's completely different. The article you pointed at was all about
porting applications between platforms. The project refered to in the
original artical was about running an existing Linux binary inside
a zone on top of Solaris, with no recompilation.
He said "hard data."
There is a reason it's called the PC,and not a dumb terminal.
There are no dumb terminals - only dumb users.
This isn't targeted at PC users. This is for (for example) the hedge fund that needs 50 machines for 8 hours, once a week, to run a complex model. This gives them the power they need for a fraction of the price of the raw hardware, and they don't have to pay anybody to maintain it.
I've had projects where I really wanted 1000 CPUs for a week, just so I could do scalability testing. There's no way we could afford $1,000,000 to buy 1000 machines just for that one test, but we could probably have swung $50,000 to get them for five 10 hour days or ten 5 hour days.
If they're smart, the Linux guys won't be looking at it - just like the Solaris engineers shouldn't be looking at Linux. And the FreeBSD guys shouldn't be looking at anything.
With all the different licenses floating around, there is too much risk of getting polluted and accidentally mixing IP.
Oh, wait...they did that years ago.
It's just you. And whoever posted the story. According to their latest 10Q filing, they have $6.7B in current assets (i.e., cash, securities, accounts receivable, etc.).
The review is one page long. There are two paragraphs that list new features, with damn close to zero explanation of what they actually are, and absolutely no indication that the author even tried them. There is no discussion of the benefits of those features, how well they work, how easy/hard they are to use, what the performance implications are, what applications the reviewer tried, or anything.
The review states:
Unfortunately it's at this point that the Solaris proposition starts to lose some of it lustre. Yes, you can download and install it just like Red Hat or SuSE Linux, but there the similarities end, making Solaris 10 far less of an obvious choice for companies looking for a Linux alternative.
What does that even mean? What "similarities" between Solaris and Linux is he looking for and what benefits do those similarities deliver to the customer? How does the absence of these unspecified similarities reduce the "lustre" on Solaris "proposition"? This may be the single dumbest sentence I've ever seen in a review of any product.
To begin with, it's important to understand that you're still dealing with a proprietary OS here.
And?
He then goes on to complain about the Linux compatibility feature's poor emulation. It's not clear how he is able to test this, since he admits that it's not even shipped as part of the product yet.
Let's give him the benefit of the doubt and assume that he isn't just making shit up, and that he actually does have super-special access to software that Sun hasn't shipped. Maybe there is a reason Sun chose not to ship that code yet? Why is the shipping product being criticized for the quality of code that was deliberately left out of it?
This review is just a shoddy piece of work. ZDnet should be embarrassed to have their name on it and Slashdot should be embarrassed that one of their editors believes that this is a "comprehensive" review.
Or when the music starts downloading immediately (and quickly) did you think they were encrypting the 5 MB song on the fly for every download?
That's exactly what they should be doing, if they care at all about the integrity of their DRM system.
Another option is to pre-DRM and cache the songs (with the number of pre-computed copies driven by the popularity of the song), and then provide one of those pre-DRMed copies and the appropriate key to you when you buy it. Then they can refill the cache in the background.
Of course, if they don't care about the integrity of the system, then their scheme is completely acceptable. But, I doubt the music industry that provides Apple with their product would be happy to hear that the DRM scheme was never intended to be more than just window dressing.
What the RIAA should really want the DRM to do is...make sure that breaking it is enough of a PITA (or seemingly dangerous) so that the normal user won't bother.
I agree, but that doesn't seem to be what Apple has done. If they are really sending you completely unprotected files, then "breaking" the DRM isn't even remotely a PITA.
Obviously the "unwrapping" portion has to be done entirely on the client - unless you limit yourself to clients that are always connected. It's the initial "wrapping" part that should be done on the server.
Any DRM scheme is vulnerable, since in the end you do have to trust the client. In this case, Apple has made the scheme embarrasingly weak by handing out un-DRMed files in the first place.
How could Apple do something this stupid?
Whether you like it or not, DRM is the cornerstone of iTunes acceptance among the music industry. Without DRM, there is no way iTunes would even exist.
The first rule of security is that the client is untrustworthy. For Apple to put all of the security of their DRM scheme on the client side is astoundingly dumb. I expected much better of them.
Xen can also use copy-on-write (and UML can too) filesystems for the common stuff between operating systems (so if you have five identical Xen domains, they can all mount the same copy-on-write fileystem), so you don't need five installs of the same OS.
Cool. I missed that. Thanks.
Actually...if Solaris does get open sourced, wouldn't it be nice to have a Xenised Solaris?
http://news.com.com/2102-7344_3-5581484.html
They would need to evaluate the market for Sun on mini-ITX platforms and conclude that it can add at least a few hundred million per year to their top line before it gets interesting
If you ever have access to Sunsolve, look at bug 6212267. It's a Priority 1 bug that prevented S10 from booting on Epia mini-ITX systems. It was marked an "S10 stopper", meaning they would have held S10 until the bug was fixed. I don't expect Sun to start shipping tiny low-power systems like this, but they obviously care about them.
1) It doesn't work well enough.
Meaning what? It's not fast enough? It's not robust enough? It's not fully-featured enough? All of those are false, but you're not even being specific about your complaints.
2) It takes away from Sun's core business, which is and always been selling Sparc-based hardware
Sun's core business has always been selling networked Unix systems and the services to go along with them. Those systems have mostly been (Ultra)SPARC based for 13 years now, but that's just an implementation detail.
3) It doesn't provide a compelling alternative to Linux in the application space which is targeted by inexpensive x86-based hardware.
It doesn't have the device support that Linux does, but it runs on a huge amount of x86 hardware. I think the performance, stability, and features like SMF, Zones, and dtrace make it a compelling alternative.
4)I've been trying to do Solaris on x86 for 10 years now, ever since Solaris version 2.3. It just isn't stable.
I stuck with Linux on x86 until S10 came out, so I can't speak to Solaris' stability on PC hardware before that. As of S10, Solaris on x86 machines is just as stable as Solaris on SPARC machines (*) and certainly more stable than Linux.
(*) OK, that's not quite true. It's as stable as Solaris on SPARC within the limitations of the hardware. SPARC systems tend to have higher-quality components, better recovery from ECC errors, multiple paths to I/O devices, and so on.
well played troll... i'll bite.
/usr/sfw.
/opt/sfw.
...
No need to be a dick. He ran into reasonable problems with some of Solaris' rough edges.
next, don't trust solaris x86 on any hardware that doesn't say sun on the outside
Bull. S10 is humming along just fine on my 2 CPU Dell, my Thinkpad T42 (modulo the Centrino-based Wifi), my homebrewed Epia file server, and my homebrewed 2-way Opteron system. Solaris doesn't have the driver support of Linux, but it still runs on a ton of different hardware.
Solaris x86 is basically a direct port from sparc
No. Solaris is Solaris. The Solaris running on your x86 machine is exactly the same as the Solaris running on your SPARC. Obviously there is some platform-specific code, but it is _not_ a port. They are built from the exact same source tree.
All non-sun software goes to
Or
Welcome to Solaris, if you don't like it, leave and keep preaching for
Again, don't be a dick. He was trying to use Solaris and ran into trouble. It happens. If you look at the rest of the thread, it's obvious that he is looking for suggestions, and is willing to try them out. Do you actually think your semi-informed arrogance is going to make anybody more interested in using Solaris?
TFA says that unlike Xen and UML, Zones have a very small overhead.
/var and /etc.
Compared to other virtualization technologies Xen has a low performance overhead, but it still isn't 0. With Zones, the performance overhead really is 0.
Was he talking about performance only, or other resources as well?
With Xen, you have to staticly partition physical memory among the domains, which can be wasteful if the domains have different workloads. With Zones, the resources can shift between zones dynamically based on usage.
With Xen, each domain has a full install of the OS, which takes quite a bit of diskspace. You can probably get around that by setting each up as a diskless client, but how many people really do that? With Zones, the bulk of the OS image is shared by all the zones, saving disk space. My local zone takes about 65MB on disk, mostly for the zone-specific files in
Just to be clear: none of this is meant as a criticism of Xen. It is a very different beast than zones, so these just reflect different tradeoffs. Xen gives you better isolation and more flexibility in OS choice than Zones do, but those benefits do come with some cost.